Dell Alienware M15 R7 (intel) LA-L651P does not recognize Nvidia RTX 3060M dGPU

[judamex]

Hello, I have a Dell Alienware M15 R7 INTEL LA-L651P which does not detect the dGPU which is a 3060M, I tried reprogramming both BIOS, without success, the dGPU has all the voltages, I do not know how to make it turn on properly with an external power supply, since outside the case, it only turns on for a second and turns off.
I tested in vBios and on pin 2 of the eeprom I have no voltage, on pin 1 I do have 1.8V
I also disabled the hybrid GPU in the BIOS settings and the Windows startup sound is heard but there is no image, with a color code of 2 red and 7 blue on the power LED. It did give me an image with the Intel iGPU.
What voltages could I measure?

[Sephir0th]

Explain what you mean with "not detected" other than the symptom of Blackscreen in non-hybrid mode.

You are absolutely sure that voltages like CORE, FBVDDQ, PEX and all the other GPU related power rails are up?

[judamex]

Explain what you mean with "not detected" other than the symptom of Blackscreen in non-hybrid mode.

You are absolutely sure that voltages like CORE, FBVDDQ, PEX and all the other GPU related power rails are up?

It doesn't appear in the device manager, in BIOS or in Dell tests. I also tried with nvflash, mats and mods and driver without success.

As far as checking all the coils are powered, but I can't check them completely, because I don't know how to turn on the motherboard outside the case, I don't know if it detects something specifically connected to turn it on completely, since if I only connect the power button, it turns on for a second and turns off.

I checked the first voltages as best I could in a few seconds, and they were all there. The only one I couldn't find in the schematic was 1.2VS_VGA. I don't quite understand where that voltage comes from.
I also don't understand what the schematic means by "GPU GC6 Entry sequence."

[judamex]

Explain what you mean with "not detected" other than the symptom of Blackscreen in non-hybrid mode.

You are absolutely sure that voltages like CORE, FBVDDQ, PEX and all the other GPU related power rails are up?

UPDATE!

After checking a few times and trying different BIOS, I managed to get video output again with the IGPU, but it still doesn't detect the GPU.
I have the schematic for LA-l651p and boardview of similar cards LA-L761P or LA-651p that help me identify some connections, but I'm a bit stuck on this topic.
I'm sure I measured 0.78V on NVVDD
0.95V on Pex_VDD

As you can see in the images, the GPU driver is not detected in the BIOS, Device Manager, Driver, or NVFlash
.

[judamex]

Update 2:
I managed to check important voltages on UG6 and UG7, but I'm missing PEX_RST#, which I don't quite understand where it's generated or why it's missing.

[mon2]

Did not read your entire thread but on PEX_RST#, this may be the PCI Express reset pin found on the PCIe expansion slot connector.

[judamex]

Did not read your entire thread but on PEX_RST#, this may be the PCI Express reset pin found on the PCIe expansion slot connector.

In summary, the Dell Alienware does not detect the dGPU, in this case the Nvidia RTX3060. It has all its voltages to work. I have checked the voltages in UG6 and I do not have PEX_RST#, what I do not know is where it comes from specifically for this board or how the resistors seen in the image work.

[mon2]

See page 7 of the schematics.

[judamex]

See page 7 of the schematics.

Does that mean the PEX_RST signal is generated by the CPU? Are we talking about a faulty CPU or one with broken balls?

[mon2]

Carefully flux and remove UG6 off the board. After UG6 is removed, configure your meter into DIODE mode.

Red meter probe, yes RED to ground. Black meter probe onto the pad that mates with this signal on the removed UG6 pcb pad. What is the measurement? If this trace is missing due to a bad solder joint, the measurement will be OL = floating state of this trace. If indeed the (hidden) ball is mating with the CPU, then you will have a voltage drop from the internal diode of the CPU. Post your measurement. You can solder back the UG6 device after this test.

[judamex]

Carefully flux and remove UG6 off the board. After UG6 is removed, configure your meter into DIODE mode.

Red meter probe, yes RED to ground. Black meter probe onto the pad that mates with this signal on the removed UG6 pcb pad. What is the measurement? If this trace is missing due to a bad solder joint, the measurement will be OL = floating state of this trace. If indeed the (hidden) ball is mating with the CPU, then you will have a voltage drop from the internal diode of the CPU. Post your measurement. You can solder back the UG6 device after this test.

The measurement is 0.835 in diode mode

[judamex]

Carefully flux and remove UG6 off the board. After UG6 is removed, configure your meter into DIODE mode.

Red meter probe, yes RED to ground. Black meter probe onto the pad that mates with this signal on the removed UG6 pcb pad. What is the measurement? If this trace is missing due to a bad solder joint, the measurement will be OL = floating state of this trace. If indeed the (hidden) ball is mating with the CPU, then you will have a voltage drop from the internal diode of the CPU. Post your measurement. You can solder back the UG6 device after this test.

What would happen if I bridge 1.8V_aon to pex_rst#?

[mon2]

Do not hard force the logic state.

Is ug6 onboard or removed? Measure the resistance to ground of this signal.

For testing, ok to move the pull down of 100k ohms on this line to the open pads to make the resistor a pull up to 1v8_aon.

If the resistance of this line is very low, the pull-up will not work.

Many articles on this topic note this and gate is the root cause of this fault.

What is the voltage of the power good of this gpu? Check the output of the other and gate below this one in post #5.

[judamex]

Do not hard force the logic state.

Is ug6 onboard or removed? Measure the resistance to ground of this signal.

For testing, ok to move the pull down of 100k ohms on this line to the open pads to make the resistor a pull up to 1v8_aon.

If the resistance of this line is very low, the pull-up will not work.

Many articles on this topic note this and gate is the root cause of this fault.

What is the voltage of the power good of this gpu? Check the output of the other and gate below this one in post #5.

The above reading is with the UG6 removed and the resistor in place. The measurement is 0.835 in diode mode.

The readings on UG7 are:
pin1: 2.7V
pin2: 3.4V
pin3: GND
pin4: 1.8V
pin5: 1.8V

Readings on UG6:
pin1: 0V
pin2: 3.3V
pin3: GND
pin4: 0V
pin5: 1.8V

Reading on pin 1 without the 100k pull-down resistor and UG6
Red Pex_rst#. Black Ground: 77k
Diode mode
Red Ground, Black Pex_rst#: 0.832

reading pin 1 without resistance and with UG6 connected

Red Pex_rst#, Black GND:74k
Diode mode
Red Ground, Black Pex_rst#: 0.738

[judamex]

Today I decided to put everything back together and run tests with Nvidia mats/mods since I was seeing that in the 12th generation onwards for some reason the graphics stop being detected due to memory failures, so I decided to run a test and this is the result.
The GPU is not detected by these tests, which means that the problem could be due to a lack of voltage or some other unknown fault. I hope this can complement my post.

[judamex]

Today I decided to put everything together and give up on this laptop, simply using it without a dGPU, which led me to install all its drivers and clean Windows, but I realized something very peculiar, I can't adjust the brightness on the screen and the HDMI is not detected for any reason. I've heard of a fault in these computers, the famous MUX, I don't know if this has to send a signal to the PC and activate the GPU.